Detecting COVID-19 infection using Chest X-rays
The exponential increase in COVID-19 patients is overwhelming the healthcare systems across the world and it continues to have a devastating effect on the population. A critical step in the fight against the virus is a fast and reliable testing technique. The conventional techniques (RT-PCR) is costly and more importantly they take time and have limited sensitivity.
While the diagnosis is confirmed using polymerase chain reaction (PCR), infected patients with pneumonia may present on chest X-ray and computed tomography (CT) images with a pattern that is only moderately characteristic for the human eye. In late January, a Chinese team published a paper detailing the clinical and paraclinical features of COVID-19. They reported that patients present abnormalities in chest CT images with most having bilateral involvement.
Detecting possible COVID-19 infection from Chest X-ray will provide a faster and reliable method that could be used with the conventional tests for faster detection. Since most modern healthcare systems are equipped with digitized X-ray machines, there are no additional costs or resources required for testing.
The Model
The system used consists of a model which has been trained on 250+ Chest X-ray images. The model uses computer vision techniques to determine the anomalies in the images and determines whether the patient has been infected with COVID-19. Most of this anomalies are invisible to the naked eye as shown below.
The data used to train the model was made available publicly by the University of Montreal. For detailed explanation on how the model works, check out this Colab.
Results
The model when tested showed an accuracy of 99%. i.e the model can effectively predict if a person is affected by COVID-19 with Chest X-rays with almost 100% certainty.
This result implies that this is a very good approach for detecting possible COVID-19 infections. The model is very accurate and with even more data, the model will be able to perform extremely well in the current scenario.
With the help from the medical community this approach could be tested out and improved as it would significantly reduce times between tests and would strengthen the healthcare systems in the fight against COVID-19.
